3D Feature Matching for Navigation Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image-based navigation systems face challenges in performing computationally efficient 3-D feature extraction and matching while maintaining accurate feature matching, especially when dealing with environments containing many identifiable features.
Innovation Solution
The system employs a method that selects segments from both image and map data, extracts features using algorithms like SIFT or SURF, and then uses two-dimensional and three-dimensional descriptors to identify corresponding features, with 3-D information from LADAR or stereo vision to prune incorrect matches and update navigation solutions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If 2-D matching algorithm is used to match features in image data, then computational complexity is reduced, but feature matching accuracy deteriorates leading to incorrect matches
Solution Approach 1:
The patent transitions from 2-D feature matching to 3-D feature matching by incorporating elevation data from LIDAR or digital elevation models. Features are matched in three-dimensional space rather than two-dimensional image space, adding a vertical dimension that resolves ambiguities in 2-D matching and improves accuracy while maintaining computational efficiency through selective 3-D processing of candidate matches.
2Reliability
If 3-D feature extraction and matching is performed, then feature matching accuracy is improved, but computational expense increases
Solution Approach 1:
The patent segments the feature matching process into distinct stages: 2-D feature extraction from images, 2-D feature matching to identify candidate pairs, and selective 3-D verification using elevation data. This segmentation allows the computationally intensive 3-D processing to be applied only to candidate matches rather than all features, significantly reducing overall computational expense while maintaining high accuracy.
Solution Approach 2:
The patent applies 3-D feature matching partially rather than universally - specifically to candidate feature pairs identified through initial 2-D matching. This partial application of 3-D processing provides sufficient accuracy improvement without the full computational burden of processing all features in 3-D space, optimizing the balance between accuracy and computational expense.
3Reliability
If all identifiable features are considered for matching, then feature matching completeness is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary 2-D feature matching to identify candidate feature pairs before applying 3-D verification. This preliminary action filters the large set of all identifiable features down to a smaller subset of promising candidates, allowing complete and accurate matching without processing time proportional to the total number of features in the scene.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Systems and methods for feature selection and matching are provided. In certain embodiments, a method for matching features comprises extracting a first plurality of features from current image data acquired from at least one sensor and extracting a second plurality of features from a prior map, wherein the prior map represents an environment containing the navigation system independently of data currently acquired by the at least one sensor. The method also comprises identifying at least one first feature in the first plurality of features and at least one second feature in the second plurality of features that have associated two-dimensional representations; and identifying at least one corresponding pair of features by comparing a three-dimensional representations of the at least one first feature to a three-dimensional representation of the at least one second feature.